System and method for striped mirror re-synchronization by logical partition rather than stripe units

Abstract

Striped mirrors are re-synchronized by logical partition rather than by stripe unit for stale partitions. While re-synchronizing physical partitions, data is serialized across both the stripe unit and the real partition region. The re-sync Read operation uses the normal logical partition offsets to read on a physical partition boundary rather than using the stripe units. In so doing, a single disk in a striped mirror can be replaced without re-synchronizing the entire mirror. The resync Read is not broken up into individual stripe unit reads.

What is claimed is:  
     
       1. A data processing system implemented method for resynchronizing a striped mirror, comprising: 
       receiving information that a disk is stale, wherein the disk is associated with a mirror; and  
       performing a resynchronizing Read for a block on only the stale disk, wherein prior to performing a resynchronizing Read, the method comprises:  
       serializing an IO request across both a stripe unit and a partition boundary.  
     
     
       2. The method recited in  claim 1 , wherein performing a resynchronizing Read further comprises: 
       determining an active mirror for the block;  
       reading the block from an active partition on the active mirror; and  
       writing the block to the stale partition on the stale disk.  
     
     
       3. The method recited in  claim 2 , wherein reading the block further comprises: 
       reading the block from the active partition as a single Read.  
     
     
       4. The method recited in  claim 2 , wherein reading the block further comprises: 
       reading the block on a real partition boundary on the active partition.  
     
     
       5. The method recited in  claim 2 , wherein reading the block further comprises: 
       reading the entire block from the active partition.  
     
     
       6. The method recited in  claim 2 , wherein reading the block further comprises: 
       reading the entire block as a single re-sync request.  
     
     
       7. The method recited in  claim 2 , wherein writing the block further comprises: 
       writing the block to the stale partition as a single Write.  
     
     
       8. The method recited in  claim 2 , wherein writing the block further comprises: 
       writing the block to a real partition boundary on the stale partition.  
     
     
       9. The method recited in  claim 2 , wherein reading the block further comprises: 
       writing the entire block as a single re-sync request.  
     
     
       10. A data processing system implemented method for resynchronizing a striped mirror, comprising: 
       serializing IO requests to both a logical region boundary and stripe units  
       issuing a re-sync Read IO request for a block;  
       determining an available mirror for the block;  
       issuing a single Read for reading the block from a boundary on a first disk, wherein the first disk is associated with the available mirror; and  
       issuing a single Write for writing the block from a boundary on a second disk, wherein the second disk is stale.  
     
     
       11. A data processing system implemented method for resynchronizing a striped mirror, comprising: 
       issuing a Read IO request for a first block;  
       determining at least two stripe units associated with the first block;  
       identifying a physical region for each stripe unit;  
       issuing separate Read IO requests for each stripe unit associated with the block for each physical region;  
       issuing a re-sync Read IO request for a second block;  
       determining an available mirror for the second block;  
       issuing a single Read IO request for reading the second block from a boundary on a second physical region, wherein the second physical region is associated with the available mirror; and  
       issuing a single Write IO request for writing the second block from a boundary on a third region, wherein the third region is stale.  
     
     
       12. A data processing system for resynchronizing a striped mirror, comprising: 
       receiving means for receiving information that a disk is stale, wherein the disk is associated with a mirror;  
       performing means for perform a resynchronizing Read for a block on only the stale disk; and  
       serializing means for serializing an IO request across both a stripe unit and a partition boundary.  
     
     
       13. The system recited in  claim 12 , wherein the performing means for performing a resynchronizing Read further comprises: 
       determining means for determining an active mirror for the block;  
       reading means for reading the block from an active partition on the active mirror; and  
       writing means for writing the block to the stale partition on the stale disk.  
     
     
       14. The system recited in  claim 13 , wherein the reading means for reading the block further comprises: 
       reading means for reading the block from the active partition as a single Read.  
     
     
       15. The system recited in  claim 14 , wherein the reading means for reading the block further comprises: 
       reading means for reading the block on a real partition boundary on the active partition.  
     
     
       16. The system recited in  claim 13 , wherein the reading means for reading the block further comprises: 
       reading means for reading the entire block from the active partition.  
     
     
       17. The system recited in  claim 13 , wherein the reading means for reading the block further comprises: 
       reading means for reading the entire block as a single re-sync request.  
     
     
       18. The system recited in  claim 13 , wherein the writing means for writing the block further comprises: 
       writing means for writing the block to the stale partition as a single Write.  
     
     
       19. The system recited in  claim 13 , wherein the writing means for writing the block further comprises: 
       writing means for writing the block to a real partition boundary on the stale partition.  
     
     
       20. The system recited in  claim 13 , wherein the reading means for reading the block further comprises: 
       writing means for writing the entire block as a single re-sync request.  
     
     
       21. A data processing system for resynchronizing a striped mirror, comprising: 
       serializing means for serializing IO requests to both stripe units and a logical region boundary;  
       issuing means for issuing a re-sync Read IO request for a block;  
       determining means for determining an available mirror for the block;  
       issuing means for issuing a single Read for reading the block from a boundary on a first disk, wherein the first disk is associated with the available mirror; and  
       issuing means for issuing a single Write for writing the block from a boundary on a second disk, wherein the second disk is stale.  
     
     
       22. A data processing system for resynchronizing a striped mirror, comprising: 
       issuing means for issuing a Read request IO request for a first block;  
       determining means for determining at least two stripe units associated with the first block;  
       identifying means for identifying a physical region for each stripe unit;  
       issuing means for issuing separate Reads for each stripe unit associated with the block for each physical region;  
       issuing means for issuing a re-sync Read IO request for a second block;  
       determining an available mirror for the second block;  
       issuing means for issuing a single Read for reading the second block from a boundary on a second physical region, wherein the second physical region is associated with the available mirror; and  
       issuing means for issuing a single Write for writing the second block from a boundary on a third region, wherein the third region is stale.  
     
     
       23. A computer program product for implementing a method for resynchronizing a striped mirror, comprising: 
       receiving instructions for receiving information that a disk is stale, wherein the disk is associated with a mirror;  
       performing instructions for performing a resynchronizing Read for a block on only the stale disk; and  
       serializing instructions for serializing an IO request across both a stripe unit and a partition boundary.